Speaker: Kang Li (UCF, Mathematics Department)
Title: A Kernel Distance Covariance Reliant Fréchet SDR method
Abstract: We present a novel Fréchet sufficient dimension reduction (SDR) method using kernel distance covariance to handle metric space-valued responses (e.g., functionals, count data, probability densities). By mapping non- Euclidean responses into a kernel feature space, our approach achieves efficient and flexible dimension reduction. Through simulations and real applications, it outperforms existing Fréchet SDR techniques, underscoring its robust and broad applicability for complex data.
Read More